Solved!
Go to SolutionUnity Replication and ESXI
We have 2 Unity arrays; UnitySource and UnityDestination
We've created the replication ports and session to replicate data from Source to Destination. All is green and seems fine, data moves to destination just fine.
When we initiate a planned failover, ESXI loses access to the Datastore. I assumed there was a function in Unisphere that handled routing requests from the Source to the Destination but I can't seem to make that happen.
When creating the replication session the Destination is named the same as the Source by default. I thought this was to help with failover but that doesn't seem to be the case. VMware still tries to communicate with the Source but fails since the failover command makes it read-only.
Anyone been able to make this work?
DELL-Sam L
Community Manager
•
8056 Posts
•
34380 Points
0
0
Posted February 5th, 2024 19:21
Hello UCJL,
What is your current OE version? When you configured your unity did you use Unity Family Configuring Replication guide? https://dl.dell.com/content/manual80453175-dell-emc-unity-family-configuring-replication.pdf?language=en-us
Thanks for the link Sam.
This is slightly different then the one I used but I went through the steps here for testing.
Software Version: 5.3.1.0.5.008
Management IP addresses for Destination and Source are on a dedicated network segment.
Replication interfaces for Destination and Source are on their own dedicated network segment.
Management and Replication networks can talk to each other.
Created a test datastore in Storage->VMware called TestRep
Created the Replication Session->
Source-> TestRep Asynchronous Destination-> TestRep
I assumed the system names both datastores the same by default as part of the replication process in case of a failover but I could be mistaken there.
I put a test VM on datastore Source->TestRep and 15 minutes later the data now resides in Destination->TestRep
However, if I initiate a failover to Destination->TestRep VMware (ESXi 7.0.3) loses access and the VM dies.
I'm missing something about this config but can't pinpoint it.
I know it's a slightly older thread, but that's expected behaviour. When you do a planned failover, you need to shutdown the VMs on the datastore, remove access to the source datastore, then do the failover. After that you give the hosts access to the target datastore, and re-import the VMs. It will recognise the target datastore as a snapshot, since it is a different device with the same signature.
DELL-Sam L
Community Manager
•
8056 Posts
•
34380 Points
0
0
Posted February 7th, 2024 16:22
Hello UCJL,
I would suggest that you open a support case so that we can look at the logs to see what is going on and how best to resolve this issue for you.